9512.net
甜梦文库
当前位置:首页 >> >>

Biomimetic Robots for Shallow Water Mine Countermeasures 1


Biomimetic Robots for Shallow Water Mine Countermeasures
1
Joseph Ayers*, Jan Witting*, Cricket Wilbur*, Paul Zavracky**, Nicol McGruer** and Donald Massa*** Marine Science Center*, Dept. of Electrical and Computer Engineering**, Northeastern University Boston and East Point, Nahant, MA 01908 Massa Products Corporation*** 280 Lincoln St., Hingham, MA 02043.

1

Supported by DARPA through ONR Grant N00014-98-1-0381

1

Abstract
We are developing two classes of biomimetic autonomous underwater vehicles based on animal models with superior performance in shallow water. The first is an 8-legged ambulatory vehicle, that is based on the lobster and is intended for autonomous mine countermeasure operations in rivers, harbors and/or the littoral zone ocean bottom with robust adaptations to irregular bottom contours, current and surge. The second vehicle is an undulatory system that is based on the lamprey and is intended for remote sensing operations in the water column with robust depth/altitude control and high maneuverability. These vehicles are based on a common biomimetic control, actuator and sensor architecture that features highly modularized components and low cost per vehicle. Operating in concert, they can conduct autonomous investigation of both the bottom and water column of the littoral zone or rivers. These biomimetic systems represent a new class of autonomous underwater vehicles that may be adapted to operations in a variety of habitats.

Biomimetic Robots
One of the hopes of the biomimetic approach to robotics is that we will be able to capture the performance advantages that animals systems enjoy in the natural environment. Marine animals such as lobster and lamprey operate in environments that challenge both human divers as well as robotic implementations. Recent advances in microcontrollers, smart materials, microelectronic devices and the understanding of animal behavior have made it feasible to build truly biomimetic robots. We report the development of two types of vehicles that may capture the performance advantages of lobsters and lamprey in the littoral zone, harbors and estuaries. These robots represent an attempt to design robotic systems around the organizational principles and capabilities of animal systems. The technologies we employ include:

?

Biomorphic Plant: Animals have a form that confers stability in the environment that they occupy. An example is the low splayed posture of the robotic lobsters, which affords considerable stability in the roll and pitch planes (Fig. 1). Our robotic implementations include a tail and claws to provide hydrodynamic stability. Myomorphic Actuators: We have implemented force recruitable artificial muscle using a smart material: Nitinol (Duering, et al., 1993). Neuromorphic Sensors: Animal sensors utilize a labeled line code which consists of sensory modality (light, pressure, etc.), magnitude of the stimulus and orientation of the stimulus relative to the plant (receptive field). We have realized this form of coding using MEMs sensors. Behavior Libraries: We base the behavior of the robots on a library of behavioral sequences reverse-engineered from movies of lobsters behaving under the target conditions (Ayers, 2000). Neural Circuit-based Controller: Animals have evolved a conservative network organization for motor systems based on command systems, central pattern generators and coordinating systems (Stein, 1978; Pinsker and Ayers, 1980). We have implemented this architecture using object software techniques to mediate the behavioral libraries described above (Ayers and Crissman, 1992). 2

? ?

? ?

The advantage of this approach is that many of the components are conserved among phyla. Thus these elements form a library of components which may be assembled into a variety of different robot types based on animal models which occupy different target environments. We have been using this neurotechnology to develop lobster, lamprey and ultimately scorpion-based robots (Ayers, et al., 1998).

Figure1. Lobster Robot.

Ambulatory Robot The first vehicle is a legged autonomous underwater ambulatory robot (Fig. 1). The vehicle is designed for operation in shallow waters that feature current and surge. It consists of an 8” by 5” hull actuated by eight three-degree of freedom legs and stabilized by anterior and posterior hydrodynamic control surfaces that extend 8 inches in front of, and behind the hull. It is powered externally or by a rechargeable battery and is controlled a neuronal-circuit based controller which implements a behavioral set reverse engineered from the action sequences of lobsters adapting to the candidate environment. The behavioral set is based on commands organized around a set of state variables that define the action of the robot. These state variables define the direction and intensity of locomotory movements as well as the postural state of the other control systems such as body height, pitch and roll. The actuators use pulse width modulation of current to achieve graded contractions. These gradations of intensity can range from activity of different muscles to the overall amplitude of a behavioral act. The system is relatively fault tolerant and can operate when several legs and/or muscles are inoperative. The system can receive higher order commands from a remote operator using a sonar transponder system (Massa et al., 1992). This vehicle is intended for remote sensing, mine countermeasures in the littoral zone of the ocean and/or rivers and streams. The vehicles may be delivered in torpedo cache systems or from small craft to an area to be investigated and cleared. The overall scenario is for the robots to achieve a comprehensive search of a mine candidate area using a swarm of autonomous vehicles. We hold that the set of behavioral acts which a lobster employs to search for food is exactly what a mine hunting 3

robot needs to perform to localize and identify mine candidates. Thus the behavioral set of the vehicles will be derived directly from behavioral reverse engineering of the behavioral sequences of real lobsters (Ayers, 2000). The robots will rely on sonar as well as near-field queues, which are primarily tactile, in the localization of mine candidates. By using swarms of the vehicles, the probability of collision with mine candidates is maximized while the number of missed collisions is minimized. The vehicles will operate as supervised autonomous agents in a landing lane delineated by a stand-off acoustic lane marking system being developed as a Phase II STTR program with Massa Products (Fig. 2). The role of the lane marking system is two-fold (1) to constrain the search within the candidate landing zone and (2) to supervise the search activities to insure complete coverage. A master controller will supervise the movement of the vehicles and maintain a record of the tracks of the vehicles. The controller will use this record to constrain the movements of the vehicles, to insure a complete search pattern, although at the level of the individual vehicles the search will be random. As the operation proceeds the supervisory system will influence the vehicles to insure complete coverage of the landing lane. The landing lane will be delineated by a set of sonobuoys that are equipped with high frequency transponders. During initiation of a search, the vehicles will initiate a search segment on an arbitrary heading, then annunciate its position. When a vehicle annunciates, the master controller will use the acoustic cues to localize the position of the vehicle. When the position is determined, The vehicles will be given a heading and an arbitrary length of time to locomote on that heading. At the end of that time period the vehicle will annunciate completion of that segment of the search. This search segment procedure will repeat to allow comprehensive coverage. As the search proceeds the record of search paths will be used to determine regions of low coverage and the controller will on a probabilistic fashion direct search segments in those regions. This algorithm assumes that the behavior of the robot will be autonomous during a search segment and that the competencies of the vehicle will allow it to deal with environmental contingencies. These contingencies might include uneven substrates, rocks, boulder fields, shoals, wave surge, tidal currents, algal beds, etc. The vehicle will have the overlying motivation to navigate on a specified compass heading. When it encounters obstacles it will attempt to ascertain whether the obstacle is a mine candidate or not. There are numerous types of sensors which can be used for identification that are under development including electronic noses, acoustic hardness testers, active electric field perturbations which may be used in identification. The investigative behaviors adopted from lobsters will insure that the sensors can be both brought into adequate proximity as well as deployed from all orientations relative to the mine candidate. If the vehicle determines that the obstacle is not a mine candidate it must make the decision whether to climb over the obstacle or to transverse around it. The vehicles will use a combination of antennal sensors and claw like surfaces to ascertain whether climbing is feasible. Where climbing is unfeasible the vehicle will use a wall following algorithm to locomote around the obstacle until it can resume its predetermined heading. This basic scenario will apply to almost all sea floor types. Where the substrate slopes the vehicle will rely on orientational reflexes to maintain stability in the pitch and roll planes. On the basis of tactile queues from leg sensors the vehicle will be able to determine whether the bottom is cobble, sand or hard. Segmental leg reflexes will enable the vehicle to adapt its stepping movements to cobble bottom substrates. The vehicles will be able to extend their claw like surfaces forward to allow them to plow through algal beds.

4

Figure 2. A. Landing lane delineated by sonobuoys. B. Detail of search with ambulators exploring the bottom while the undulators operate in the water column.

Operation in the littoral zone or rivers will require adaptations to current and wave surge. When wave surge is high and especially when the rate of change of current is high, the vehicles will evoke a sequence of rheotaxic behavioral acts that will serve both to streamline the vehicle as well as insure hydrodynamic stability. These rheotaxic acts will include yawing into the current, pitching the body forward, lowering the claw-like control surfaces and elevating the tail-like control surfaces. Each of these components of rheotaxic behavior will have a defined threshold as perceived by current and shear sensors. During investigation the vehicle will rely on its capabilities for omnidirectional locomotion to circle around the object while maintaining sensor suites on the object at close proximity. By performing this procedure, from different orientations the vehicle can use voting algorithms to fuse information about the size, shape, surface properties, and electrical, magnetic or chemical signatures that it may use for identification. At the completion of the mission the vehicles can be commanded to deploy a miniature lift bag to force them to float to the surface where they can be collected for reuse.

Undulatory Robot

Figure 3 Undulatory Robot

5

The undulatory robot (Fig. 3) is intended to complement the operation of the ambulatory robots to perform search operations for mine candidates that are suspended in the water column. It will be delivered similarly to the operational theatre by torpedo cache systems or small craft and will be governed in its search behavior by the same lane marking system as the ambulatory vehicle. During searches the vehicle will navigate on search segments which are specified by a compass heading. The vehicle will use a sonar altimeter to regulate its altitude to that of suspended tethered mine candidates. While swimming on an arbitrary heading, the vehicle will use high frequency directional sonar to insonify the water in front of the vehicle and listen for sonar returns that indicate a close range object. Since the vehicle will be undulating in the yaw plane, the sonar will be scanning laterally. If the vehicle retains the compass heading at which it generated the sound pulse, it can use omnidirectional sonar receivers to correlate the reception of an echo at short latency with the orientation of the transmitter when the pulse was generated. By this mechanism the vehicle will be able to localize the orientation of candidates relative to the vehicle. When a short latency return is obtained, the vehicle will begin investigative behavior. To investigate a candidate the vehicle will perform turns toward the candidate to maximize the probability of collisions with the candidate as well as maintain station on the candidate. Where possible, a set of hooks will be embedded in the anterior end of the vehicle to allow it to tangle into any fouling material on the surface of the mine. If the vehicle becomes entangled it can deploy sensors which can aid in identification or simply turn on a sonar beacon to identify the location of the candidate.

Myomorphic Actuators
Antagonistic pairs of muscle elements that either operate joints (Ayers and Davis, 1977; Ayers and Clarac, 1978) or flex the body axis (Ayers, 1989) mediate the walking and swimming movements of animals. We have developed low level pattern generating circuits based on neurophysiological models. These circuits generate patterns of actuator activation which mimic those generated by motor neuron pools (Fig. 4, upper panel). The control signals activate artificial muscle formed from the shape memory alloy, Nitinol (Duering, 1993). Individual muscle elements are formed from a single strand or loop of nitinol that is insulated from the seawater by a Teflon sheath. The nitinol is stimulated to contract by a train of current pulses that heat the wire above a transition temperature that causes the wire to transform to a compact crystalline structure (austenite) and as a result shorten by up to 5%. When the current ceases, the wire is cooled by the surrounding water and can be mechanically stretched by the antagonistic muscle to a deformable state (martensite). These contractions can be graded in amplitude by current pulse width modulation (Fig 4., lower left panel). Changes in the pulse width duty cycle recruit different proportions of martensite to austenite and allow graded contractions. Like animal muscle the velocity of shortening in this artificial is decreased when the load increases. In the lobster each muscle is controlled by a few motor neurons which are organized in order of increasing size which corresponds to their force producing capabilities (Davis, 1971). We mimic this arrangement by having a set of motor synergies that evoke current pulse trains of increasing duty cycle. Thus to achieve weak, intermediate and strong contractions we recruit these synergies in order of increasing size (Fig 4).

6

Figure 4. Myomorphic actuators. Top Left Panel: Motor patterns generated by the ambulation controller to mediate forward walking (left) and lateral walking on the trailing or pushing side (right). Lower Left Panel: Current pulse width modulation circuit utilized to generate graded contractions of the nitinol actuators. Lower Right Panel: Three degree of freedom walking leg based on nitinol artificial muscle.

These artificial muscle modules are used to actuate biomorphic leg systems in the lobster (Figure 4) or a polyurethane body axis in the undulator. In both cases contractions by one muscle mechanically causes the transition from austenite to martensite in the antagonist.

Neuromorphic Sensors
Animal sensors code environmental parameters in terms of labeled lines (Bullock, 1978). Each sensor is represented by an array of labeled line elements, each of which codes for a particular sensory modality (gravity, water current, etc) as well as a receptive fields (i.e. orientation relative to horizontal, water currents from the front, rear or the sides, etc.). In the controller implementation, sensor objects pass messages to command objects based on the actual input. Thus the message contains the modality and orientation and evokes different methods from the command system, based on these parameters (Ayers, 2000). During locomotion in currents and surge, lobsters predominately rely on three sensors, the antennae, water current receptors and the statocysts or balance receptors (Hoyle, 1976; Kennedy and Davis, 1977). Although the actual biological sensors are 7

complex, they can be readily modeled by mechanical transducers to code environmental information in the same fashion as the lobster nervous system (Bullock, 1978).

Figure 5. MEMs-based Antennal Sensor. A. Structure of the high dynamic range bending sensor. B. Increase in dynamic range achieved by more bendable base and stop system. This arrangement minimizes sticktion by mechanically forcing the switch open after a deflection. C. Mechanical drive used to control the antennal structure in the yaw plane of the vehicle. D. Arrays of MEMs switches are used to achieve range fractionation of imposed deflections. E. Final Antennal device with MEMs sensors distributed along the shaft of the antennae.

Antennae: Real lobsters use their antennae extensively both as tactile receptors for collision avoidance as well as water current receptors. We are implementing antennae using bend sensors that are mounted on a flexible substrate (Fig. 5a-b, d). The switches are organized in an array so that as the magnitude of the imposed bend is increased, different switches are closed. The antennae are constructed of a tapered polycarbonate laminate of moderate restoring force. A motor system mediates protraction and retraction in the yaw plane so that the antennae can be deployed at different orientations as well as waved to feel for objects (Fig. 5c). MEMs switches placed at the distal end will code for contact while others placed at the basalar portion code for water currents as well (Fig 5e).

8

Figure 6 MEMs –Based Flow Sensors. A. Scanning electromicrograph of an array of flow sensors on the silicon substrate. B. Oil-filled cap system used to insulate the switch from the surrounding seawater. C. Claw control surface with the MEMs sensor sequestered in a depression to protect it from collisions with rocks, etc.

Flow Sensors: The ability to compensate for flow and surge requires directional flow sensors (Fig. 6). Real lobsters mediate this sense with both their antennae as well as with specialized sensors called hair fan organs. We are implementing a MEMs flow sensor based on the hair fan organ. The device consists of a MEMs switch that features a vertically projecting flap that projects into the flow (Fig. 6a). Flaps of different length allow estimation of the magnitude of flow by closing a different pattern of switches. The switch itself is covered with a cap that is filled with oil to protect the switch from the seawater (Fig. 6b). The flow sensors will be mounted in depressions on the claws to protect them from damage and oriented at different angles relative to the midline to determine the direction of imposed flow.

Behavioral Libraries
Finite State Analysis of Lobster Behavior We have developed computer controlled video technology for reverse animation and kinematic analysis of animal behavior (Ayers, 1989, Ayers, 1992). This multi-media system allows correlated acquisition of kinematic and electrophysiological data by simultaneously recording behavior in the video signal and electrophysiology on the audio channels of a high-resolution digital VCR. We developed extensions to a public domain image analysis program (NIH Image) which include the capability for color based acquisition and image segmentation as well as time-based quantification of kinematic parameters and correlated analog acquisition (ColorImage, Ayers and Fletcher, 1990; Ayers, 1992). This system allows us to measure animal orientation, joint angles from video on a frame by frame basis to establish the detailed movement strategies kinematics of compensatory, orientational and taxic reflexes as well as the underlying neuromuscular control signals. As a result it has been possible to establish the coordination patterns and control signals 9

underlying omnidirectional walking (Ayers and Davis, 1977) as well as undulatory swimming (Ayers, 1989).

Figure 7 Examples of states used in behavioral quantization.

The assumption of this analysis is that the posture and action of the different task groups is specified by a set of command systems (Bowerman and Larimer, 1974a,b; Kupfermand and Weiss, 1978) that command the task group to generate a different state (Fig. 7). Thus the task of this analysis is to specify the state of the task groups in each of the frames of a movie. We have found the following set of states adequate to define the ongoing behavior of both the lobster and the robotic vehicle.

?Thorax Pitch: rostrum up, level, rostrum down. ?Thorax Roll: left up, level, right up ?Thorax Yaw: hard left, easy left, straight, easy right, hard right ?Thorax Height: high, normal, low ?Walking direction: standing, forward, backward, lateral leading lateral trailing ?Walking speed: slow, medium, fast, stop ?Claw Pitch: up, normal, down ?Claw Yaw: (left and right): extended, normal, meral spread, lateral spread ?Antennae Yaw: (left and right): protracted, normal, lateral, retracted ?Uropod Posture: flared, normal, adducted ?Abdominal Pitch: extended, elevated, normal, depressed, flexed
Our analysis of the sequencing of these task groups borrows from a technique utilized by astronomers to detect motion of galactic objects. As the analysis proceeds through each frame of the digital movie, the program flashes between temporally adjacent frames of the movie with a brief pause after each cycle. Appendages that are moving the most flash in these projections. A panel of buttons that represent different states of the task groups (e.g. elevation vs. depression of the claws, etc.) are available to the investigator to specify which groups are active. By clicking on the appropriate buttons for each frame, it is possible to efficiently quantify the activity of all task groups at high temporal resolution from videotapes of specimens behaving in a variety of situations. These state tables are used to establish control sequences for the robots based on the behavior of the model organisms. 10

Fig. 8. Finite state Analysis of animal behavior. Screen shot ot the application ColorImage during analysis of a sequence of lobster behavior. Upper left panel: Digital movie of lobster behaving in an aquarium. Lower Left Panel: Radio-button panel used in estimating the state of the different task groups from the movie. Lower Right Panel: Cartoon of the robot as it would appear with the state selection for comparison with the animal in the movie.

We have completed development of the behavioral analyzer, and integrated the output of the analyzer with the controller (Fig. 8). This program unifies the behavioral analysis with the mechanical control problem by providing a state-based simulation of how the real robot will respond to the behavioral control signals. It also resolves the major bottleneck in the behavioral analysis, namely the ability to assess the accuracy of the behavioral decomposition and to readily recognize behavior subunits. A major feature of this player is that It defines completely the states that the real robot will need to implement to achieve the performance capabilities of the animal model and will allow us to design behavioral acts de novo as well as reverse engineer acts from behaving animals. These states sequences are stored in the form of state matrix tables, where each row of the table corresponds to one video frame and each column consists of one of the state of one of the variables described above.

11

Figure. 9 Postural changes underlying rheotaxic sequence.

Behavioral Controller
The controller for these robots is organized at two levels. At the lowest level a pattern generating circuit generates the pattern which controls individual legs in the ambulator or segments of the body axis in the undulator. The organization of this pattern generating system has been described (Ayers, et al. 1998). At the highest level the controller is organized around the behavioral libraries and a sequencer. The sequencer maintains the state of the vehicle by switching between different command states in a temporal sequence specified by both orientation and magnitude of sensor input and the response patterns specified by the behavioral libraries. During operation, the vehicle is given an underlying motivation (i.e. home on a sonar beacon or initiate a search on an arbitrary compass heading). During execution of this motivated task, the controller will respond to environmental contingencies specified by its sensor suite by evoking preprogrammed sequences from its behavioral library. . An example of such a behavioral sequence is rheotaxic behavior or orientation and compensation for flow or surge (Figure 9). Adaptive responses to water currents involve simultaneous compensation in the pitch plane an increase in the walking forces, as well as postural responses of the tail and claws. The ambulation controller supports an antigravity recruiter that acts 12

on the depressor synergy of each leg and mediates pitch and roll compensation. An example of compensatory responses of the controller to flow directed toward the side of the specimen is indicated in Figure. 9. During response to surge sensed by the MEMs flow sensors, the controller first lowers the body, spreads the claws and orients into the current. It then reduces the depression in anterior segments which will pitch the hull forward, extends the claws and elevates the tail.. These compensatory responses to water currents and surge involve both yaw and pitch plane components as well as load compensation for the necessary added propulsive thrust. Software Objects of the Behavioral Sequencer The sequence of commands that underlie ongoing behavior is maintained on an event stack (Fig. 10). Action components are pushed on this stack in order of time by releaser objects. Releasers are triggered by specific messages from sensor objects. The primary advantage of this arrangement is that compatible action patterns can be superimposed and the layers corresponding to different task groups can be managed in parallel. Releaser objects are triggered by different methods of sensor objects that indicate orientational and magnitude components. Sensor objects receive input from an array of MEMs sensor switches and activate different releasers depending on the character and receptive field of the sensors activated by exteroceptive input. A releaser (1) saves the current state of the system, (2) queues the times of parametric modulation events (see below), (3) queues the times of commands and transitions and then returns to the event loop to pop transitions off the queue. A releaser can thus activate multiple command objects in parallel, orchestrating a complete behavioral act. In many cases, behavioral acts which operate in parallel superimpose upon each other at the level of the effectors (von Holst, 1973). In other cases, especially when presented with the releasers for two incompatible behavioral acts, the animal typically chooses to perform one act over the other and one behavioral act suppresses the other. In our controller suppressor objects, triggered by releasers mediate behavioral choice. These objects embody lateral inhibitory connections between commands and are the locus of implementation of behavioral choice (Edwards, 1991) Suppressor objects for a particular behavioral act maintain a look up table of action components that the act suppresses and then clear incompatible action components from the event stack..

Figure 10 Command stack organization of the behavioral sequencer.

13

Modal action patterns are triggered by releaser messages from sensors in an all or none). In the controller, they consist of a list of action components that are placed in sequence on a queue. The components specify transitions of action, posture and intensity. We are expanding this list based on empirical observations from reverse kinematics of behaving lobsters (see above). Goal achieving patterns have both a releaser which triggers the action pattern sequence as well as a terminator which ends the. During execution of the action pattern the system can switch between modal action patterns based on sensor feedback. This process continues until the terminator message is received from the sensors.

Navigation The yaw components of navigation are mediated reactively by taxes and kineses (Loeb, 1918; Braitenberg, 1984). Positive yaw taxes or attraction occur when sensor bias directs locomotion toward source and are generally mediated by contralateral causality between sensors and effectors (Fig. 11). Negative yaw taxes or avoidance occur when sensor bias directs locomotion away from source and are generally mediated by ipsilateral causality between sensors and effectors (Fig. 11). In the ambulation controller, such inputs send messages to both the parametric command as well as the recruiting component. In attractive reflexes the messages is sent to the contralateral command objects while during avoidance reflexes (negative taxis) the messages are sent to the ipsilateral command objects. As a result, the controller both biases the period on the two sides (walking faster on the side turned away from) as well as the recruiters on the two sides to generate more propulsive force on the faster side. Control schemes like that of Fig. 11 can be easily adapted to a variety of beacon tracking, 14

Figure 11. Causal networks for taxic behavior.

avoidance, docking and search behavior through acoustic, magnetic, optical, tactile, gravitational, flow and chemical sensors. Modulation of Motor Patterns Many external sensory inputs modulate the intensity of motor output or superimpose a taxic component modulating yaw or orientation (Kennedy and Davis, 1977). Modulation is realized by modulator objects that sequence modulatory actions on the event stack (Fig. 10). Modulator objects can be divided into trigger components and gate components (Stein, 1978). Trigger components mediate modulation that persists longer than the releasing stimulus. Trigger modulation is characterized by (1) Rise Time, (2) Duration, (3) Fall Time, (4) Amplitude. Gate modulation, in contrast, persists only as long as the stimulus that releases. In our controller the intensity of behavioral acts is controlled by the temporal compression used to place them on the queue as well as by variations in the the pulse widths associated with different recruitment levels which will modulate the amplitude of the behavior.

Conclusion
It is feasible to build truly biomimetic robots that have the potential to capture the performance capabilities and behavior of their animal models. These implementations feature a conservative control architecture that can be adapted to a variety of habitats. The behavioral sets that the animal models employ in their search for prey provide a mission script which is equally applicable to the search for underwater mines. Their high maneuverability makes them capable of placing sensor suites in close proximity to distinguish mine candidates from actual mines. Use of smart materials may reduce that cost of these systems to the point where they may be use on an exchange of assets basis. References
Ayers, J. (2000) Finite State Analysis of Behavior and the Development of Underwater Robots. In Artificial Ethology, D. McFarland and O. Holland [eds]. Oxford University Press. In press. Ayers, J. and F. Clarac. (1978) Neuromuscular strategies underlying different behavioral acts in a multifunctional crustacean leg joint. J. Comp. Physiol. 128: 81-94. Ayers, J. and Crisman. J. (1992) The Lobster as a Model for an Omnidirectional Robotic Ambulation Control Architecture. In: Biological Neural Networks in Invertebrate Neuroethology and Robots., R. Beer., R. Ritzmann and T. McKenna [eds], 287-316. Ayers, J. and W. J. Davis. (1977) Neuronal control of locomotion in the Lobster, Homarus americanus. I. Motor programs for forward and backward walking. J. Comp. Physiol. 115: 1-27. Ayers, J and Fletcher, G. (1990) Color Segmentation and Motion Analysis of Biological Image Data on the Macintosh II. Advanced Imaging 5: 39-42 Ayers., J., Zavracky, P., McGruer, N., Massa, D., Vorus, V., Mukherjee, R., Currie, S. (1998) A Modular Behavioral-Based Architecture for Biomimetic Autonomous Underwater Robots. In: Proc. of the Autonomous Vehicles in Mine Countermeasures Symposium. Naval Postgraduate School., In press. Bowerman, R. F. and Larimer,J.L; (1974a) Command fibres in the Circumoesophageal connectives of crayfish I. Tonic fibres. J. Exp. Bio. 60: 95-117. Bowerman, R. F. and Larimer, J.L. (1974b) Command fibres in the circumoesophageal connectives of crayfish II. Phasic fibres. J. Exp. Biol. 60: 119-134. Braitenberg, V. (1984) Vehicles: Experiments in Synthetic Psychology, MIT Press, Cambridge. Bullock, T. H. (1978) An Introduction to Neuroscience, Freeman, San Francisco. Davis, W. J. (1971) Functional Significance of Motor Neuron Size and Soma Position in the Swimmeret System of the Lobster. J. Neurophysiology 34: 274-288. Davis, W. J. (1979) Behavioral hierarchies. Trends in Neurosciences : 5-8. Davis, W. J. and J. Ayers (1972) Locomotion: Control by positive feedback optokinetic responses. Science 177: 183-185 Davis, W. J. and Kennedy, D. (1972) Command Interneurons Controlling Swimmeret Movements in the Lobster 1. Types of Effects on Motoneurons. J. Neurophysiology 35: 1-12. Davis, W. J.; Mpitsos, G.J.; Pinneo,J.M. (1974) The Behavior Hierarchy of the Mollusk Pleurobranchaea II. Hormonal Suppression of Feeding Associated with egg-laying. J. comp. physiol. 90: 225-243. Duering, T. Melton, K. N. Stockel, D. and Wayman, C. M (1990) Engineering Aspects of Shape Memory Alloys. Butterworth Heinemann, Ltd. London.

15

Edwards, D. H. (1991) Mutual inhibition among neural command systems as a possible mechanism of behavioral choice in crayfish. J. Neurosci. 11: 1210-1223. Evoy, W;. and Ayers, (1982) Locomotion and Control of Limb Movements. In: The Biology of Crustacea, Vol. 4. Neural Integration and Behavior. D. C. Sandeman and H. Atwood, [eds]. Academic Press, New York, Pp. 62-106. Hoyle, G.[ed]. (1976) Identified Neurons and Behavior of Arthropods. Plenum Press. Kennedy, D. and Davis, W.J. (1977) Organization of invertebrate motor systems.In: Handbook of Physiology. Geiger, R., Kandel, E. Brookhart, J.M.[eds]. American Physiology Society. Bethesda Md,. Pp. 1023-1087. Kupferman, I. and K. R. Weiss (1978) The command neuron concept. Behav. Brain Sci. 1: 3-39. Loeb, J. (1918) Forced movements, tropisms and animal conduct. Lippincott, Philadelphia. Pinsker, H. M. and Ayers, J. (1983) Neuronal Oscillators. Chapter 9 in: The Clinical Neurosciences. Section Five. Neurobiology. W. D. Willis [ed]. Churchill Livingstone Publishers. Pp. 203-266. Stein, P.S.G. (1978) Motor Systems, with specific reference to the control of locomotion. Ann. Rev. Neurosci. 1: 61-81. von Holst, E. (1973) The Behavioral Physiology of Animals and Man. University of Miami Press, Coral Gables.

16


赞助商链接

更多相关文章:
更多相关标签:

All rights reserved Powered by 甜梦文库 9512.net

copyright ©right 2010-2021。
甜梦文库内容来自网络,如有侵犯请联系客服。zhit325@126.com|网站地图